In Germany's pharmacies (and all around the world), you can buy a tooth paste with "pearls". These are little, soft grains, and the combination of the seemingly suggested concept "sandpaper" (they even have to deny that on the front page of the product's home page) with the clean and beautiful image of "pearl" seem to lead people to believe that these particles will make your teeth twinkle in the dark. I am not in the capacity to judge whether those actually perform better deplaquification. The tooth paste goes by the name of "Pearls and Dents" (for non-German speakers). That's another German product carrying a name made of English words, which appears to be cool these days. I find it a sad development how English replaces German all over the German media and every day language use (second only to the germanification of English words, like "downgeloadet" -- the use of English words according to German grammar rules). But that's not the point, at least not this time. What I find even more disturbing is the stupidity of the masses that go with the flow, or even try to be creative at designing captive slogans, regardless of the widespread lack of command of the English language. "Dents" may be close to "dental", but it does not actually mean teeth. Thus I am left to wonder whether the manufacturer is trying to hint at what those little pearls actually do to your teeth, or whether it's just another instance of a crap marketing team (who'd probably be better off learning proper German first, before venturing into the domain of another language). Along similar lines, is it only me who thinks that Adidas' slogan "impossible is nothing" is just plain stupid. Are they trying to be extra clever by giving the all too common phrase "nothing is impossible" a witty twist (and failing at it)? Or are they instead hinting that "impossible" is for others -- wear Adidas and you'll encounter even less possible situations? PS: This post is dedicated to helix.

Rapha l Hertzog and Roland Mas, both well-known Debian developers (and Rapha l is also a Debian book author, which Roland has proof-read), have done it: my book is now also disponible en la langue de l'amour! Thanks a lot to both of you for all your work (and sweat (and blood (and, well, there weren't any tears i hope...) -- nor blood)). I cannot wait to get my own copy! As with the Japanese translation, it feels weird to hold a text in your hand with your name on it but without a chance to actually read it. My French actually isn't that bad, so I am looking forward to browsing it, but as with Junichi and Kenshi, the Japanese translators, I feel quite comfortable about the translation, having met Rapha l and Roland in person and having witnessed their dedication to Debian and thoroughness of their work. So again: thanks a lot, guys! In addition to German and Japanese, as well as the original English, this is now the forth language, and there are more to come. PS: you wouldn't believe it, but the French edition with its 674 pages tops all other editions in terms of length: it's 25 pages longer than the German one. Update: I forgot to mention that all of the aforementioned translators, Junichi, Kenshi, Roland, and Rapha l have taken their time to send feedback and corrections to me. That deserves another round of thank yous!

Julien has reason to be proud. But his sysrqd webpage delivered a clear stab to my eye:

[...] Authentication is made by clear password.

I really thought those days were over! Having a clear-text protocol with single-password authentication may make sysrqd responsive under high load, it'll also ensure that I'd never touch it. I think it should be possible to implement at least digest authentication or even PAM without being too heavy on the system.

Dear GNU: could you please just be Unix once in a while and work the way a reasonable person would expect it to work? Not the freakin crackhead way your fathers seem to prefer?

I went back to the shop today to return the Nokia E60 I had previously sought out. There were just too many gripes with the phone, and even though I could have possibly worked around (some of) them, I was looking for a phone that works, not one that I could tweak to work. Among the most prominent problems were:

• Lack of STUN support, which means that you cannot do VoIP behind a NAT router without configuring a port-forward (which is often not possible). Nokia is also unlikely to add STUN support because the phone's target audience is clearly large businesses equipping their employees with phones that will work with VoIP locally, and "fall back" to GSM when on the road.
• Broken WPA2 implementation. I could not get my phone to associate with my WPA2 router. Reasons may be the AES implementation, which seems broken (see below), or just a faulty WPA2 implementation. I can get other devices, including my Palm, to talk to the router over WPA2 just fine.
• Broken AES (or VPN) implementation, meaning I could not get a VPN connection to either of a Cisco router, a Check Point VPN-1 gateway, a Linux host running FreeS/WAN, or OpenVPN when AES was used. I could connect to all but the OpenVPN client using 3DES.
• No WLAN reconnects. If your WLAN connection drops while you're using it, you have to exit the programme (or reassociate with the VoIP gateway) to get it back up. This sucks when browsing the web!
• Sucky battery lifetime. I was playing around with the device, using a fully-charged but two year old battery. After about 2.5 hours, it displayed the first low battery warning. Obviously, WLAN consumes a lot of power, but that's just too much and too fast.

Other than that, the phone seems to have the usual features you expect from an advanced phone that tries to also be a PIM. Believe it or not, I could make phone calls, but there was weird crackling noises to be heard; I am not sure whether the phone's at fault, I do have crazy power lines around my place. One thing that did drive me up the walls was the configuration. To set up VoIP, I had to first configure a WLAN connection (which you cannot do in the "Connection Manager" -- that one seems to be only used to kill connections and associate with unencrypted WLANs) in the system settings (which are to be found under Menu/Tools), then set up a SIP profile, then set up a VoIP connection, and then make sure it's active. If you change WLANs, you have to modify the VoIP connection, it cannot use an existing WLAN connection, or the current best default. I want a VoIP/GSM combo not because I want to spend two minutes to configure VoIP in a foreign network. Also, while we're on a topic related to WLAN: the device does not allow you to connect to a WLAN without starting an application that uses it. Sure, it's power-optimised, but still. Also, neither ESSID nor encryption status are shown anywhere when you're connected, and I did not find out a way to display the phone's IP address. Another beef was with the general device speed. There's a quick access button to write SMS. If you press that, it takes 5-6 seconds until you can type the SMS. I don't need to say more... The display of the E60 (and E70), however, do warrant their own paragraphs: it's gorgeous! Simply gorgeous. Unfortunately, that's not a selling factor for me. So I decided to wait for the next generation of WLAN/VoIP/GSM combos (or at least firmware updates to the E60), and especially the Motorola A910, which runs Linux. I am grateful to the shop manager for being so patient, making sure that I am satisfied rather than trying to make money, and generally just giving me a pleasant shopping experience the way it really should be these days -- given that there are shops out there that won't let you try electronic devices before you buy them. So, if you're ever in Zurich and in need of mobile communication devices or paraphernalia, visit the mobilezone shop in Oerlikon (Einkaufszentrum Neumarkt) and ask for Andreas Iseli! Apart from being next door to my flat, I'm definitely going to go back there next time I need something. PS: to all those who have been toying with the idea of buying (or otherwise drooling over) the E70, do note that the two run the same software, so most of these issues will apply to the E70 as well. However, it's always best to make up your own mind, so find a shop as professional and fair as mine and get your hands on it.

Ubuntu caused a lot of friction with and for Debian. In discussions with its founder, Mark Shuttleworth, and other Ubuntu developers during (and before) Debconf6, I was able to spell out the main criticisms from the Debian perspectives of the way Canonical/Ubuntu is handling things (without a claim to completeness). These criticisms mainly stem from discussions with fellow developers over the past 18 months, and I largely support all of them. I am publicising them here to help make the status quo more transparent. Before I go on, I want you to know that my neutral and somewhat even supportive stance towards Ubuntu has led many Debian folks to conclude that I must be a Canonical employee; I never was nor will be. Let's go:

• A major point of friction I've gathered from many (Debian) sides stems from the fact that many developers are largely unsatisfied with how Canonical/Ubuntu cooperates and "gives back" to the Debian community: essentially, at the moment all we get are monolithic patches describing the differences between Debian packages and their Ubuntu counterparts. Even though Mark has claimed that the patches should be modular if the underlying package uses a modular package management system (such as dpatch), this does not seem to be the case at first glance (c.f. e.g. squid). You have to look inside the monolithic patch to note that its changes are compartementalised into dpatch files. Yet, Canonical/Ubuntu continuously claims to be "good" and to "give back" to the Debian community, but that's just not happening in that way. Canonical/Ubuntu either has to start giving back according to their own claims, or stop pretending that it does, and it's good that I have heard this voiced by one of the Canonical employees as well.

• The perception that Canonical/Ubuntu is taking advantage of Debian seems to prevail. Taking advantage in this definition entails using but not giving back, as well as dragging integrative work out of Debian, something Joey Hess described as the supermarket thing, which has the potential to ruin the Debian project. Whether this is true or false, intentional or unintentional, it is important to recognize that these sentiments do exist among a number of members of the Debian community. There's a related point, namely jealousy about Ubuntu's success, and the fear that it steals Debian's users. In response, it is often said that while Ubuntu depends on Debian, the reverse is not the case: Debian does not need Ubuntu. This is false and true at the same time. Debian surely need derivatives like Ubuntu, for derivatives bring users, and users bring improvements and weight. However, Debian does not need any one derivative more than any other, and in fact it has been considered harmful if any derivative stands out.

• Many of those with whom I've spoken perceive that Canonical/Ubuntu do not acknowledge Debian enough, nor give credit where credit is due. And this lack of credit and acknowledgement further amplifies the fear. Two examples seem to stand out:

  • The "universe" community that gathered around Ubuntu's core surprisingly often does not know about Debian, or do not know how Ubuntu and Debian relate. I do not have any data on this, unfortunately (see below). This is something to take up with the Ubuntu Community Council.

  • On the ubuntu.com front page, Debian is not mentioned at all; if you actually follow the About Ubuntu link, you can find Debian mentioned once in the first paragraph (albeit without a link), and if you did notice the right hand menu expanding, you may even decide to click on Ubuntu and Debian, where the relationship is finally illustrated from the Ubuntu point of view. To many, this is just not enough. Update (+6h): the About Ubuntu page now links to the Debian homepage. Thanks!

• Canonical people, and Mark especially, draw an important line between Ubuntu (which is mostly free/open-source), and Canonical with its other products (Launchpad, Rosetta; which are non-free). However, from the outside, this line does not exist, or at least it appears as if there's a much stronger relationship between the Canonical and the Ubuntu project. This impression casts a corporate shadow over Ubuntu which deters many developers, even if it's undeniable that Ubuntu is a project of its own (to a large degree).

• There's no question about Mark being on an agenda, for which he has assembled an impressively powerful team (which indirectly includes many Debian developers) at an (even more impressively) cheap cost. It's not my business to speculate about other people's finances, but those that do conclude that Mark is very likely not running too much of a risk with what he does at the moment and could basically pull out at any point in time. To some, who have dedicated large parts of their lives to Debian, this seems like an unwelcome way to play a game.

• When Canonical entered the market and hired some developers to work on a Debian-related project, a lot of jealousy boiled up among those who didn't get a job, because back then it seemed that Canonical was out to pay people to work on Debian -- which is a common dream among us developers; at least people hoped that's what it would be. Many fundamental contributors felt left out and confronted with the question why they should continue their work for free when others are now getting paid for it. The twist is that nobody wanted to ask themselves that question, because money was never the reason why they started to work on Debian. The situation has leveled by now (I think). More and more people realise that Canonical is taking Ubuntu into a direction where many Debian developers do not want to go. Nevertheless, it seems that some of the frictions we are currently facing between Debian and Ubuntu still have their roots in the bad feelings from those times (which may well still exist in the individual case). Canonical/Ubuntu (and some Debian developers) need to realise (and get over) this, and somehow I think they've got a long way to go.

On Friday night of the Debconf6 conference, Mark assembled a bunch of people discuss the inter-project relationship. The (large) number of people who eventually joined, as well as the choice of venue (with live music) made the discussion somewhat awkward, but we did manage to get some good talking done, and I raised all of the above points. See the minutes of the meeting if interested. With their latest release out the door and a little more time on their hands (at least for a while), I hope that Canonical will follow up on the resolutions of the evening, as they have promised. The same goes for Debian, of course. There's very little Ubuntu can do about the jealousy other than acknowledge it, but it would be an important start. In addition, I hope to see all of the following take place rather sooner than later:

• quality improvements to the patches they "give back" to Debian, and the adoption of a more appropriate medium of exchange. Instead of patches, active reaching out to the Debian developers by the Ubuntu developers would be even better in certain cases. The claim by Mark and one of his employees that this is simply impossible because of the commonly misconceived size of the paid body of developers (which is just too small) is really their challenge, and if they cannot meet it, maybe they have to go around it. It should be noted that efforts along these lines exist, so Canonical does not have to start from scratch. Note to Debian developers: if you're so inclined, please add your name to the bottom of that page.
• a clear statement about how Canonical/Ubuntu plans to coexist or cooperate with Debian, and then the execution of those plans without leaving any doubt.
• a survey of the Ubuntu community to gather (and publish) information about common perceptions of Debians position and role with respect to Ubuntu and vice versa. Again, this is something for the Ubuntu Community Council.
• steps to ensure better familiarity with the Debian-Ubuntu situation among the Ubuntu community members.
• a dedicated space for Debian people to go and be heard. Maybe the #debian-ubuntu channel on irc.debian.org will suffice, if some of the Canonical/Ubuntu higher-ups make an effort to keep an eye on it.
• further steps by Canonical towards improving the relationship between the two projects, which could include regular meetings (or even irregular ones), uniting prominent representatives from both sides.

If I may add a thought that has come up lately (it's not only mine): a derivative like Ubuntu, namely one pushing Debian to the desktop, is an important asset for Debian. However, the only way this is going to work from the Debian perspective is as a mutual effort. If Canonical/Ubuntu do not realise this, maybe another derivative has to step in to fill the void? I hope this blog entry accomplishes what it tries, namely to shed some (more) light onto the relationship between Debian and Ubuntu, and to help the Canonical/Ubuntu side better understand the Debian perspective. Sorry for the length, I've worked on this entry ever since Debconf6 and despite a lot of editing (also by others; thank you!), it just wouldn't get shorter. Thanks for reading it all (assuming that's what you did if you arrived here). Update: this post has spread significantly, it seems. Thanks to all who've passed it on. I know of the following threads that ensued:

• http://www.osnews.com/comment.php?news_id=15106
• https://lists.ubuntu.com/archives/sounder/2006-July/008113.html

If you find any others, please let me know.

I am on the look for a new cell phone, and now that the Nokia E60 and E70 are out, I went to the shop to had a close look at them, and one thing I can say up front is that they're both quite okay for people who like Series 60 (v3) phones and all the colourful navigation stuff you (have to) do with them. I personally don't really care as long as I can quickly learn to find my way, and as long as the physical aspects of navigation are okay and responsive (which rules out Sony Ericsson for me). With that I can start at one difference between the two E series phones: the navigation nipple on the E60 seems a lot easier to use than on the E70. That said, the two phones are otherwise very similar under the hood (see this link for details, in the following are only the points I consider important):

• both are triple-band with WCDMA support.
• both come with bluetooth, infrared, GPRS, EDGE, WCDMA, and WLAN (incl. WPA/WPA2 and EAP support).
• both have a VoIP client, though the store manager didn't know for sure whether I could connect to a SER (SIP express router) on the 'Net over a WLAN connection. They say that you can connect to a local gatekeeper [0], so I'd assume it's possible, but that's something to try out; read on...
• both can assign ring tones per contact. Nokia heard my cries...
• both can reuse my 6230's battery and charger. This is really a major selling point of Nokia -- all of their phones have the same charger, and batteries are compatible across many!
• both run Series 60 v3, Symbian OS 9.1, which is somewhat outdates (9.2) is current, but it does give you "improvements in the OS [which] mean that applications and content, and therefore a developers investment, are better protected than ever." (Wikipedia). Boy am I glad!
• both can run ScummVM, meaning I can play all of the best games ever released! If this is not a selling point for a business phone, what is?

Obvious differences include:

• the E60 has no camera, while the E70 comes with a 2 megapixel one. I am actually not sure whether this is a plus or a minus, given that it is nice to have a camera handy when your life isn't all cubicles and security-conscious companies.
• the E70 sports a keyboard that you can fold out (as does the 6822). Unfortunately, here in Switzerland, all they offer is a QWERTZ version, which I'll avoid if at all possible. I don't think I'll be using the phone for "working" (as in word processing or spreadsheet editing; who does?), and I avoid SMS when I can, so even though the keyboard is certainly cool qand usable, I won't draw too much benefit.
• the E70 seems to have better battery life (according to the specs), but that's something that can be addressed with a more powerful battery, I'd say (plus, I can use my 6230 battery with both).
• the E60 is 3 cm smaller and weighs 10 grammes less than the E70. That said, it's still a lot bigger than the 6230 I now have.

You can compare the specs of the two phones as well as the 6822 (which is like the E70 in some ways, but without WLAN and VoIP) at this link. Oh, and then there's the Motorola A910, which runs Linux, comes in a little heavier and at about the same volume as the E60, doesn't include infrared support, and doesn't seem to provide a VoIP client, which I'd say is less of a problem given the Linux basis. Unfortunately, it does not seem to available in Switzerland. Together with a contract extension, I can get either of the E series phones at about half the street price. In addition, the shop manager told me he'd allow me to return the phone within three days if it didn't meet my expectations. All of this, and the above combined makes me want to sleep for another night, and then try out the E60 over the weekend. I have to make a decision whether to prolong my contract by 24 months, or only 12 months and pay CHF 100 more for the phone. That comes down to around 5 /month, which is around 10% of my monthly bill. Given that I was always happy with Sunrise as provider, it would seem natural to bind myself for 24 months. However, prices are expected to drop substantially over the next year, and I am not sure whether a provider could lock you in to an existing contract, even if the government ordered a price decrease. So in the light of the expected availability of more and enhanced WLAN phones, such as the Motorola A910, I think a 12 month contract will be the better choice.

[0]

"gatekeeper" is an H.323 term and I brought it up in the conversation because I am new to SIP and, well... He said it would be able to connect to a company-internal "thing, like a Cisco router," so this leads me to assume it will be able to connect to any SER out there. Thanks to Mark Brown for bringing this to my attention.

My book is now available in Japanese (ISBN 4-8399-1897-X), thanks to the great work by our (Debian's) very own Kenshi Muto and Junichi Uekawa! It even comes with its very own Obi -- a ribbon wound around the book with commercial messages designed to draw the attention of the (potential) reader. The word derives from the sash worn with a kimono. It's weird to hold a text in your hands that has your own name on it, but which you cannot read, especially when you're such a pedantic perfectionist as I am. This dilemma I picked up in the Japanese translation's preface, which I include in full for your reading pleasure. And I'll do so both in Japanese as well as English, just because I don't know when I'll ever get the chance again to post some authentic Japanese on my blog!

Debian Debian Kenshi Muto Junichi Uekawa Debian Kenshi pbuilder Junichi Debian Kenshi Junichi honsho ga minasama no ochikara ni narereba saiwai desu.

And here for all those who cannot understand the signs from the land of the rising sun:

I have only had the chance to visit Japan once to date, but the trip left some unforgettable memories, which make it all the more an honour for me to see a Japanese translation of my book on the market. The Debian operating system is gaining popularity fast in Japan, and it is my hope that the book you are holding in your hands will help to accelerate this growth. Please allow me at this point to express my gratitude towards Kenshi Muto, a Debian book author himself, and Junichi Uekawa, author of pbuilder, a vital component in the making of my book! Kenshi and Junichi are both renowned Debian developers and have been instrumental in the coordination of this Japanese translation. As you will notice while reading my book, I invested a lot of energy into making it a quality reference. It was thus very difficult for me to accept a text with my name on it that I could not read, as is the case with the Japanese version you are holding in your hands. However, knowing the translation in the capable hands of Kenshi and Junichi quickly put my mind to rest and I am now incredibly proud to see my work translated into the language of the fast-moving and innovative culture of Japan. honsho ga minasama no ochikara ni narereba saiwai desu.

That last sentence means "I hope the book will be helpful to you," and was translated by Shuhei, a friend who's working on his Ph.D. at Zurich's AILab. Apparently Kenshi and Junichi thought it was so cute that they simply left it as is and did not translate it. In other news, my book's also available in German (and has been since late February), the French edition is on the verge of being released by Eyrolles (no link yet), and the Chinese translation is still work in progress (also, no link yet). The book's website needs a rework to integrate the other languages, and to fix up all the shortcomings I've encountered during one year of uptime with an average of around 200 unique visits per day. It's also still using Plone 2.0.5, but I just have not had the time yet to fix it all up in a migration to Plone 2.5, mainly because I've lost synchronisation with the Plone development team over the past year. Oh well. That ToDo list just keeps getting longer...

This problem related to SpamAssassin and SASL-authenticated road warriors has bugged me for a while. Micah has a patch that alleviates the problem, but it's not in upstream, and even though upstream does plan to address the problem he raised, that's down the road. Today I figured out a solution, based on the smtpd_sasl_authenticated_header configuration option available in Postfix 2.3 and later, using header_checks to rewrite headers. In the following, lines are folded for readability. The regular expression needs to be on one line, with newlines and spaces at the beginning of the line removed:

/^Received: from ([-._[:alnum:]]+ \([-._[:alnum:]]+ \[[.[:digit:]] 7,15 \]\))
    [[:space:]]+\(Authenticated sender: ([^)]+)\)
    [[:space:]]+by (seamus\.madduck\.net) \(([^)]+)\)
    with (E?SMTP) id ([A-F[:digit:]]+)
    [[:space:]]+for <([^>]+)>; (.*)/
  REPLACE Received: from auth sender $2 by $3 ($4) with $5 id $6 for <$7>; $8

Note how the host name is left in to ensure that the right header is replaced. This will replace the problem-causing Received line with the following:

Received: from auth sender madduck@madduck.net by
  seamus.madduck.net (postfix) with ESMTP id B4CD640343F for
  <madduck@madduck.net>; Fri, 30 Jun 2006 17:50:02 +0200 (CEST)

Now SpamAssassin mentions UNPARSEABLE_RELAY, but apparently that's worth 0 points, and if I wanted, I could fix that too. Now I would like to be able to do the same for clients that authenticate with TLS certificates, but I cannot figure out how.

Mozilla Firefox bug #57342 was the first bug I encountered after I had completed my switch from Galeon (which got infected by GNOME2-itis, such that you could not change basic browser settings without gnome-control-panel). It's still a bug, but I of course understand the Firefox developer's decision that a user's request to be able to view simple text files inline (<gasp>) is a low priority improvement. I won't allow myself to make a comment about the Firefox code quality that in six years, noone has come up with a working patch for this massive challenge. Do they have to switch their browser into protected memory mode and load a DOS emulator to be able to handle plain text? Fortunately, I have since found this extension. Now my life is really much simpler.

This is an email I just sent to debian-devel:

I have done some major work on mdadm (see PTS) in experimental, which is now at version 2.5.2-1 (not all web pages reflect that yet). I still have not received enough success reports to make me comfortable enough to upload to unstable, so here's another request for testing (or else etch will ship with 2.4.1). I am 95% confident that it's a smooth upgrade, though it helps if you make sure that the output of the script /usr/share/mdadm/mkconf looks okay before you reboot. Come and find me on IRC if you need reconfirmation (see below). Previously, mdadm required a newer version of initramfs-tools, one which did not exist yet. I have addressed this issue with this version, meaning that you can install it with any initramfs-tools out there, and it will just do what it can do. Details are available in the online README, which is a little more up to date than the included /usr/share/doc/mdadm/README.experimental. Please test mdadm 2.5.2-1! and drop me a line if it works, or if it does not! The package is in experimental or available from my repository for i386 or amd64. Again, do not hesitate to bug me on IRC (#debian-devel/irc.debian.org) if you have questions or need assistance.

Update: Alec Berryman discovered a slight problem when mdadm 2.5.2-1 is used in conjunction with initramfs-tools 0.64 or earlier. Please see this post for details and solutions. The online README has been updated. I don't think this warrants a new package to be uploaded to the archive, but please correct me if I am wrong. Update: Alec Berryman helped discover another problem when the root filesystem is on LVM is on RAID. If this is your setup, please do not use mdadm 2.5 at the moment. See #375879.

Due to my physical incapacitation of the last 10 months, I have become a user of Workrave. The programme now bugs me to take a five minute break every 25 minutes of working on the computer (but it's smart about figuring out when I actually do use the machine, or another one in my network, so it's not as much of a useless nuisance as it may sounds to be). So every 25 minutes, I am being told to take a rest break. In response, I find myself doing one of three things, in decreasing order of likelihood:

• click "postpone"
• get up, wander about and keep peeking at the screen to ensure that I'd be back at the keys not any longer than 301 seconds after the popup.
• do something useful and come back to the screen when I am done.

My challenge for the times to come is to get rid of the first two points on that list.

I just stumbled over this report on the 50 most expensive cities (via Boing Boing). Zurich is down to position 9 (it was in number 5 the last time I read such a report), but that's not the only surprise -- I have not perceived London, Tokyo, and Geneva to be more expensive than Zurich, nor would I consider Zurich to have become cheaper. When I spot Hanoi and Saigon (Ho Chi Minh City) in the thirties, before Amsterdam and Munich (which doesn't even appear in the top 50), I am beginning to doubt the veracity of the list. Or well, at least I'd like to know more about how they've arrived at the ranking. Or it's been too long since my last visit to Vietnam. Yeah, that must be it. Update: Mike Hommey says:

I think what you missed is that the ranking is about comparing how cities are expensive for locals, not how inherently expensive they are. Sure Hanoi might be cheap for the European, but might not be that cheap for Vietnamese, when you take into account housing prices, etc. compared to the average salary.

I am aware of this; it's only one of the reasons why I am missing information in the study.

I am sitting on the train, returning from the first FrOSCon, a conference close to Cologne, organised by a bunch of volunteers, mostly students of Department of Computer Science of the renowned Rhein-Main-Siegen Fachhochschule (which seemed to fully support the endeavour), as well as members of the Linux / Unix User Group Sankt Augustin. I gave a presentation on my doctoral research (slides available) and otherwise spent much time talking to people about topics ranging from issues in cooperation to strategical questions on Debian, and of course my well-known involvement with the Transnational Republic. It really does seem as if my little experiment in Mexico rippled a long way and made people more conscious of the meaning of key signings. At first, I did not want to go to the conference and had even called in a week before to cancel my slot. I had signed up as a speaker in February, just as I was starting to get on top of my incapacitated wrists (and the severe muscular problems in my shoulders, which are at the root of it all) after several months of "downtime". However, I had my hopes up too high and found myself still unable to work for reasonable lengths of time following my trip to Souteast Asia, as well as before, during, and after the Debian conference in Mexico I attended. Consequently, the "first results" I had promised in the abstract didn't exist, and my only option was to give a talk along the same lines as the one I had given at FOSDEM earlier this year -- and after three years of exposure to an academic field were people seemed to do nothing else than deliver the same talks over and over again, I wasn't too keen on boring my audience. But the organisers convinced me, so I went out to buy a train ticket, and sat my butt down to make as much as I could out of what I had. In the end, I was very satisfied with the talk (for one, I did not experience a total loss of slides 10 minutes into it), which was delivered to an audience of primarily non-Debian people, and I got some valuable feedback from some of them, especially during a longer conversation with a guy from the OpenOffice.org quality assurance team. Have a look at the slides if you're interested. I enjoyed myself at night at the social event, where my favourite wheat beer "Unertl" was served, and once again met up to my reputation of being among the last to go to sleep ()nd among the first to be up, as you shall see shortly) and found my place to sleep sometime between 3 and 4 in the morning, following very interesting discussion on some of the fundamental questions revolving around Debian, Ubuntu, as well as several other software projects. "My place to sleep", by the way, was on my camping mattress outside the campus on a big meadow, which I preferred over a stuffy gym hall shared with several dozens of others. I greatly enjoyed my time under the stars until 7 in the morning, when the sun was already radiating unbearable heat (I was in a black sleeping bag), some beast had stung my finger and had caused it to swell up so much that I couldn't bend it anymore, when I remembered that I am suffering from hay fever, and two stray dogs decided that they liked me. I made my way back to the campus, got lost on the way to the shower and hence could not get one for the gym hall had been closed by the time I arrived, didn't find any wheat beer for breakfast (remember, I am from Munich), and finally could not fall back asleep in the chill-out area, which was located on the top floor with windows in the ceiling and the sun right there on the other side causing happy sparks of enthalpy and a general sense of discomfort. Discovering "Mate Ice Tea" (which is just what it sounds like), I quickly got high on caffeine and spent the rest of the day wired, running about because I couldn't stay put. At least I got to see plenty of people that way. This was the first FrOSCon and I must say that I am very impressed by the organisation team. It was not difficult to conclude that they had spent a lot of time thinking and planing and researching and learning from other events. As participant, I had the impression that everything was working according to plan at all times, I never saw any of the crew members stressed out, and the service and little details really made one feel good, as participant and probably even more as speaker. We (the speakers) had our own catering room, as well as a work room attached to it, so we could always hide away from the masses and treat ourselves to a good selection of foods and drinks, and the work room was actually among the cooler (as in temperature) places to hang out, so that came in as a welcoming refreshment. Network coverage and uptime was flawless, the connection conveniently fast, and the atmosphere generally very pleasing. I definitely hope there'll be another FrOSCon next year, and that I'll be able to attend. Thank you to everyone who helped make this conference possible.